Paste it into the formatter, or pick a command — the fastest route depends on where the JSON lives. In a terminal it’s jq .; in Python it’s python3 -m json.tool; in VS Code it’s Format Document. Every option below handles standard JSON; the last FAQ answer covers what parsing and re-printing can change.

The commands, by environment

Where	Command	Output notes
Any shell, jq installed	jq . file.json or curl -s URL | jq .	2-space indent, key order preserved
Validate only	jq empty file.json	silent on success, error + position on failure
Python (no install needed)	python3 -m json.tool file.json	4-space indent, escapes non-ASCII by default
Python, keep Unicode	python3 -m json.tool --no-ensure-ascii f.json	real UTF-8 output
Node	node -e "console.log(JSON.stringify(JSON.parse(require('fs').readFileSync('f.json','utf8')),null,2))"	2-space, re-serializes numbers
Prettier	npx prettier --write f.json	formats in place, project-style indent
VS Code	open file → Shift+Alt+F (macOS ⇧⌥F)	Format Document on .json
Browser console	JSON.stringify(obj, null, 2)	2-space string you can copy

Worked example

$ echo '{"user":"ada","roles":["admin","dev"],"active":true}' | jq .
{
  "user": "ada",
  "roles": [
    "admin",
    "dev"
  ],
  "active": true
}

The three moves people actually want

Minify (the reverse): jq -c . file.json, or python3 -m json.tool --compact. One line, no whitespace — what you want for .env-style transport or a single-cell config field.

Sort keys: jq -S . sorts recursively; --sort-keys does the same in json.tool. Handy for diffing two config files where key order is noise. The formatter has a Sort keys toggle that does the same recursively in the browser (apart from whole-number keys, which always come first in numeric order) — useful when the JSON shouldn’t leave the machine.

Find the error, not just “invalid”: every CLI above tells you a position when parsing fails, but for a 200 KB payload the formatter’s error panel — exact line, column and a caret under the offending character — is usually faster than binary-searching a file. The companion piece on the 12 common JSON syntax errors shows what each error message actually means; JSON vs JSON5 explains why a config file with comments isn’t valid JSON in the first place.

Frequently asked questions

What's the fastest way to pretty-print JSON in a terminal?

jq . file.json — or pipe it: curl -s URL | jq .. It prints with 2-space indent and keeps key order. No jq? python3 -m json.tool file.json is installed almost everywhere (4-space indent by default; --indent 2 or --tab to change it).

How do I minify (un-pretty-print) JSON on the command line?

jq -c . file.json outputs the whole document on one line. The Python equivalent is python3 -m json.tool --compact file.json. Both parse the data and print it again rather than just deleting spaces, so duplicate keys collapse to the last value and some numbers are re-spelled (jq 1.7 prints 1E5 as 1E+5; Python prints it as 100000.0, and 1e400 as Infinity, which isn't valid JSON).

How do I validate JSON — not just format it?

jq empty file.json prints nothing on success and a line/column error on failure. python3 -m json.tool file.json doubles as a validator — it exits non-zero with a position on bad input. In the browser, paste it into the formatter to get the exact error line, column and a jump-to-error button.

Why does python -m json.tool escape my non-ASCII text?

It defaults to ensure_ascii, so é becomes \u00e9 — valid, but ugly. Pass --no-ensure-ascii to keep real UTF-8 characters in the output. jq keeps them natively.

How do I sort all the keys alphabetically?

jq -S . file.json sorts object keys recursively (array order is untouched). In Python it's python3 -m json.tool --sort-keys file.json. Or use the formatter's Sort keys toggle — it sorts keys at every depth too, except that whole-number keys such as "2" and "10" always come first, in numeric order (jq and Python put "10" before "2").

Does pretty-printing change the data?

Depends on the tool. This site's formatter parses your JSON and prints it back with JSON.stringify, so it is not a whitespace-only rewrite — and neither are jq, Python or Node. Ordinary data comes out with the same values, just re-indented; in this formatter (as in Node) these cases don't: integers beyond ±9,007,199,254,740,991 can lose digits (12345678901234567890 → 12345678901234567000), so keep big IDs as strings; numbers past the 64-bit float range become null (1e400); numbers are re-spelled (1.0 → 1, 1E5 → 100000); whole-number keys ("2", "10"; no leading zeros, up to 4,294,967,294) move to the front in numeric order; and duplicate keys collapse to the last value — RFC 8259 says names should be unique, so fix the source. If you need every token kept as written, use a whitespace-only formatter such as VS Code's built-in Format Document — and check signatures and checksums against the original bytes, never a formatted copy.